A factorial experiment to investigate naturalistic factors affecting smartphone gait analysis

Gait analysis involves analyzing data from body-mounted sensors in order to detect various ailments or aging. Prior work has utilized the accelerometers on Smartphones to collect gait data but required subjects to firmly attach the phones to their torsos, hip or other body parts. In real life, subjects prefer to place their phones in various pockets or bags, may wear loose or tight clothes which have been found to affect the quality of the accelerometer data gathered. In this paper, we report on factorial experiments to investigate what naturalistic factors affect accelerometer data gathered from unattached Smartphones. For a female subject, we found that the most impactful factor was wearing hard vs soft shoes. For a male subject, the most impactful factor was whether the phone was carried in an attaché case or not. Overall, we also found that none of the factors investigated produced an overwhelming response, which suggests that high-fidelity health assessments of various ailments could possibly be performed using gait data gathered in naturalistic settings (no need to attach smartphone precisely on subjects).